South Indian Tamil Language Handwritten Document Text Line Segmentation Technique with Aid of Sliding Window and Skewing Operations

In document image analysis, Text line segmentation is one of the key components. The segmentation logic presents essential information about skew correction, zone segmentation, and character recognition. The method of document image segmentation into text lines for printed text has seen numerous contributions from fellow research scholars, yet there is scope for tremendous improvement. The key challenges for handwritten document are due to writer movement, the inter-line distance changeability and incoherent distance between the components that may differ. These may be directly by segments, or curved. The area of handwritten segmentation has seen few models; very few of the research paper are proposed for Text line skew segmentation model and hence the stimulus of handwritten south Indian languages. Consequently, a better text line segmentation technique for south Indian Tamil language is proposed in this paper. The processing of Tamil language is very crucial factor because the Tamil letters are in crucial shapes and it is harder to segment the touching lines and letters from the Tamil image documents. The challenges present in Tamil language process and the existing text line segmentation methods has been improved by our proposed method, which utilizing two major techniques namely, sliding window and adaptive histogram equalization. Our proposed text line segmentation technique initially performs the preprocessing process and these preprocessed document images are given to the adaptive histogram equalization. During the histogram equalization process, the document images text characters are enhanced to view the characters more accurately. The enhanced image text lines are segmented by utilizing the sliding window operation. For accurate line segmentation, the skewing operation is performed on the line segmented result images. The implementation result shows the effectiveness of proposed technique, in segmenting the handwritten text lines from the input document. The performance of the proposed technique is evaluated by comparing the result of proposed technique with the conventional text line segmentation technique. The result shows that our proposed technique acquires high-quality text line segmentation DR, RA and F-Measure values for the number of testing documents in comparison with the conventional technique.